Juniper Brings Intent-Based Networking to the Edge

Transcription

welcome to zcast everybody i'm zscaravel from zk research as always i'm your host i'm joined today by mike bushong vp of cloud ready data center at juniper networks mike why don't you say hi to everybody hey thanks for having me today and hello everyone out there yeah so we're gonna be talking uh data center which uh you know it's a pretty exciting topic there's always lots of stuff going on in that area before we get started though i do want to give a shout out to e-week my media partner all seacasts are done in conjunction with the ewiki speaks program so thanks to e-week uh mike now you work and live and eat and everything in data centers um what's going on with data centers right we've had this massive number of people work from home we've moved more and more applications to the cloud and so i think there's some conventional wisdom out there that data centers are becoming less relevant but uh you know tell us why that's not the case yeah i think the the whole data center thing um when cloud came in right the whole idea was the cloud is going to wipe through and take out all of the data centers that everything essentially moves to amazon microsoft google oracle whoever i just that hasn't been true right i think what most people are looking at is kind of these hybrid environments certainly workloads are moving to the cloud i don't want to say that i'm like a one of the people who says that cloud isn't happening it's absolutely happening but i think what's what's really going on is there's a bit of a bifurcation in the market um for companies that are maybe at the lower end they could they might look at cloud first or cloud only because their needs are you know fairly constrained right they've got a kind of a well understood set of applications for everybody else they've got kind of a bunch of different applications and the whole idea that merely lifting and shifting an application from on-prem into the cloud was going to yield some huge financial impact that hasn't really happened and so it's left people kind of in this environment where they've got you know the you know what they've got on prem they've got you know new emerging you know let's say sas workloads and they're leveraging cloud and what this leads to is not just you know sort of these hybrid environments but but hybrid and multi environments right on-prem public public the net of all of that is that you know data center all of a sudden is no longer about the four walls that you're kind of managing and the equipment that you have in it it's kind of this operational domain this administrative domain that extends out across different different you know areas in the network and then with all of that that makes data center hot again because it's all about data center operations i think we're actually going through a bit of a data center renaissance frankly and then you mentioned the work from home thing right you get more people consuming more applications trying to do more things in a distributed way it's like that's actually built up a bunch of demand for the services that are being offered at a data center so it's not just that operations and innovation is alive it's that kind of the current environment that we find ourselves in it's really stimulating growth in this overall space then we see that when you start looking at you know kind of how companies are doing who are in this this data center proper area i like what you framed it up because data center maybe is now become a term to describe operations versus a physical place uh but i do think that you know that uh you know i think we analysts and you know i write for a lot of public media tend to over swing the pendulum sometimes we say everything's moving to the cloud or everything's moving mobile or whatever and that's just not true um i do i will say though that every business i talk to wants a cloud model that doesn't necessarily mean public cloud in fact my research shows that 95 of companies will adopt a hybrid cloud there are just certain scenarios where things will run and by the way the 95 numbers enterprises so i do think you're right small businesses can go all cloud but when you talk about larger enterprise they have specific workflows data sovereignty issues they got regulatory issues compliance issues and you just can't move everything to uh the public cloud in fact somebody asked me the other day do i see a day where everything's in public cloud and my answer was maybe but i'll long be retired so i don't care when that happens now the other big disruptor in data centers has been this thing that we call edge and this has given rise to what i call uh and gardeners in terms of distributed cloud which describes an operating model to your point where you have one logical cloud environment that spans public clouds private clouds and edge locations and that's fundamentally different because historically what you had with data centers well they were centralized centers of data right but now you've got these distributed environments so can you you talk a little bit about edge and how that's really evolved the role of the network and data centers um yeah so if we take cloud just generally i mean cloud is is you know i think it's really an operational construct um it's how do you how do you manage things the whole idea of moving to different consumption models and and you know being able to spin up applications when and where you need we look at you know the kind of classic cloud it was the idea that we would bring users and data to the application right things were hosted centrally and then you know you could sort of reroute everything into these central locations what we see with distributed edge is the kind of changing that a little bit let's take the applications let's take the data and push it closer to the users in some cases it could be because there's you know performance requirements or latency requirements it could be um you know things like 5g are actually changing the whole expectation of user experience and and what users are demanding and so it's no longer sufficient to just route everything through these national data centers what you've got to do is essentially push the data centers out closer to the users now this is actually not that different and i'll say spiritually right architecture it's very different but spiritually it's similar to this whole idea of scale out right instead of having you know these big massive things what i want to do is create smaller components and pull them and sort of distribute the load across a larger number of elements when we do that right we trade off the performance piece um we get performance advantages but what we pick up the cost of that is essentially you know distributed operations it's all the administrative things that you have to do to manage these devices well it's true in data center as well if i move away from a large national data center and i replace that with dozens you know maybe hundreds of of you know distributed data centers and this distributed edge what i pick up is an operational burden and so what we see with with the whole move to distributed edge is it's a it's a shift and it's emphasizing once again the operational tenants of cloud which brings us back to kind of the you know the opening comments around cloud it's how do i get cloud-like operations so i can reliably and repeatably deploy applications to preserve a consistent user experience when my applications and the domains over which i have to manage them are distributed in all kinds of different physical locations um with a you know diversity of connectivity a diversity of applications like that's really the game in this whole distributed edge space yeah in fact i would argue when you talk about operations a lot of businesses haven't really moved to true cloud operations i think when you think about if i pick up a legacy workload and i do a lift and shift and public cloud to a public cloud i maybe change the location of that but i'm not really changing the way i operate right so i don't get the benefit of the scale out and things like that and so this you know now what i actually think is we're actually moving to an era where we do have cloud operations and that's giving rise to true multi-cloud and i use the word true multi-cloud because historically when people did multi-cloud they were using multiple centralized clouds but to me that's not the same as multi-cloud where i could have one operational construct across all my clouds so uh you know i do think the environment does become um more you know a little more you know complicated here and so that certainly changed the role um you know of operations and security and things now um you recently made another a number of announcements right uh you know the uh through uh uh the abstract position that that juniper made can you can you just walk us through those uh yeah so i mean just as a brief reminder of what abstra does right so this whole idea of abstract is intent based networking um it's the idea that rather than working configuration statement by configuration statement you ought to be able to declare top down here's what i want my my data center to do so we did we took that intent model and we've extended it actually to the edge so we've said that you know there are architectures that are different than these centralized data centers right typically called collapsed fabrics where you you don't necessarily need a whole bunch of devices use a couple of access switches maybe some some tours um and what we're doing then is essentially extending the benefits of intent-based networking to those types of of deployments this is big right as we start thinking about you know smaller data centers uh use cases like iot we start looking at uh things like telco cloud where you've got you know let's say carriers that are you know distributing their data centers out in support of their 5g services there's a whole different kind of collection of things that we can cover within that that basic space um that's that's a big part of this overall announcement right extending the goodness of abstract to new deployments um beyond that we did some stuff on the security side which we can talk about in a bit um and then of course you know when you pick up something like intent based networking it's really is a different operating model which raises a question especially like in your enterprise clients right it raises a basic question you know there's this thing that i want that is different than what i have how do i get there and so we put together some migration services that are really aimed at how do we transition from environments as we know it today to environments as as we want them tomorrow yeah i wouldn't drill down the security side because juniper obviously has a very big uh security business as well you know when you when you think of a traditional data center it was a closed environment you could control a lot of things uh just by locking down the walls right so now that we're moving more and more things to the edge uh we're using public clouds we're kind of mixing public and private um that that really makes curing a lot more complicated so can you you talk about those challenges and then tell me what you know how juniper's addressing those sure i mean you just spelled it out if i've got different types of users accessing different types of applications all run in different types of data centers that are all supported by different types of vendor equipment and different types of vendor software like what's the problem um i mean that like like that's that should be easy it should be it sounds amazing right like what that sounds like to me by the way is job security so for all of you out there like that's you know good good stuff um i think the real challenge is you know and this is when you talk about true multi-cloud i mean you mentioned it it's the idea that i want uniform operations and i want diverse underlying you know i guess infrastructure what apps just trying to do is provide unified operations i want to do things in a common way regardless of the underlying infrastructure regardless of the vendor regardless of the the particular domain type regardless of the underlying you know users and applications whatever let me provide kind of a consistent way of applying policy across this diverse environment what we're doing in this particular release is we're taking this idea of group-based policy the idea that we have policy and control that's applied at like a tenant or an application level and then we can and essentially what apps does is it abstracts that policy from the underlying devices then we go and apply that policy across whichever devices are are required in order to support that that overall policy structure and then we'll validate that the policy that's deployed is essentially um working as intended and the way we do that right so so underpinning this idea of intent this this declarative model this way of pushing let's say configuration and state information into devices i'm underpinning that we actually create a model we say here's the the network as it's intended and then we deploy a bunch of sensors and we say here is the network as deployed and when the as intended and the deployed match we give you a big green light everything is good when as deployed and um and as intended uh do not match then we can flag that there's a problem so let me give you kind of an easy example you pay you uh you create policy for a particular application but of course multiple applications run across these different you know on the uh the switches the data center fabric what if somebody else makes a different policy instruction that overlaps with the policy for for application a and application b what we want to do is flag when there's overlapping policy we want to be able to tell you when something is different than as intended so that when you get in this complex environment of you know many to many for everything we're trying to do let us tell you if there's if there's conflicts so you can resolve that conflict ahead of time the worst time to to possibly find out that your policies are not working as intended is in the middle of your ransomware attack right so segmentation becomes a a key mitigation you know activity when you think about some of the security stuff going on you can't afford to find out that it's not going to work at the point of at the point of attack at the point that something's gone wrong what aperture's doing is not just provisioning it but telling you up front you know is your policy assured is it working the way you expect it to work uh thanks for that answer mike uh i know one of the big value propositions of appstra and there are other intent-based solutions but i believe appstore is the only real multi-vendor one so can you can you talk a little bit about why and how that's grown in importance obviously you know everybody wants to be able to choose the vendor they use but i think historically haven't been able to so what have you seen there from your customers and how they've been leveraging that sure um i mean so yes the idea behind intent based is that if we can work from a declarative model over the top then we're abstracted naturally from the underlying devices this means that we are in a very good architectural position to support multi-vendor why does that matter um look everybody is these days talking about supply chain if you want to um if you want to be able to order devices from vendor a or vendor b or vendor c but your operations are sort of fixed then you're in a position that you can't you actually can't do that what app store allows us to do is to say you know here's my data center the way it's constructed i don't care about the underlying devices my teams are trained my services are deployed applications are deployed um regardless of what the underlying vendors are then depending on what the lead times are then you can go and take advantage of you know vendor a vendor b vendor c now look supply chain won't always be a constraint right we we think that you know that things will improve at some point but at that point then the question is do you want your vendors to compete for your business if you want economic leverage for instance then keeping your vendors competing for for incremental capacity is a is a huge thing that that gives you um flexibility right it could be that they're competing based on feature functionality it could be based on you know price and performance but this gives you the flexibility to make the decisions that you need to make um you know so you can avoid these you know seven nine ten year type engagements where you say like i'm committed longer than i'm going to be in the role i'm committed to this particular vendor what we're essentially doing is unlocking that and we've seen huge traction here i mean we have um we've got folks who come and they say look you know appreciate that juniper has hardware however i've got cisco or arista or dell or sonic or you know or cumulus or juniper whatever i've got different suppliers you know can you solve my operational challenges and we've got accounts today where it's it's literally it's it's you know it's pure software over the top breathing additional life into an existing investment which allows them to get more out of what they've already spent and then gives them a plausible path to go and extract additional value over time yeah it's intended in the whole shift of software that took place in the network industry years ago was supposed to allow better network uh vendor interoperability and it never really led to that in fact i've had people describe uh their vendor lock-in situation as being a little like hotel california where you can check out but you can't ever leave and uh so you know unless you wait the best time to do that there is always with cio turnover right so that can be long uh cycles now um i i did want to ask you about the acceptance that you're seeing with intent-based networking and autonomous operations i know when apps are first launched when i when i talked to it people about this concept of autonomous operations to use the technical term it gave them the heebie-jeebies i mean a lot of them were very concerned about it and uh so what do you what do you see now and i'm assuming you know work from home has played you know a role in in getting to look at your product but are you seeing more acceptance of it and what are some of those low-hanging fruit use cases where if people were to adopted they could take advantage of it faster yeah i think people in general are getting more comfortable with automation i still think it's a long journey you know the processes that surround automation are sometimes immature especially in some of the enterprise type accounts so i think there's there's still work to go and and move forward but in terms of like basic use cases i think you start looking at how quickly can i stand up you know additional capacity how do i expand how do i add you know new racks how do i add new servers how do i add new pods depending on the size of the data center you start looking at this whole idea of mean time to innocence we have a good customer in the retail space where they have an online shopping platform and at the end of each day they've got you know there's items that sit in the um in their in their shopping carts that don't get closed and of course the the uh the the teams come back and say you know we didn't get we didn't convert these to sales there's a network problem these things timed out you know what's the mean time to innocence they spend their time literally every day you know having to prove that there wasn't some network issue that caused these purchases not to go through what we can do on the abstract side is by by demonstrating that the network is working as intended and and that there's no issues it's like that mean time to innocence is merely at that point it's like a dashboard view you know here things are working as intended that's a that takes an operational practice that would have been manual triggered by people forced to forcing other people to go in run a bunch of commands and prove that things are working as you want them to now that becomes just part of the system you know it's it's this idea that that we can provide information provide insight um like that that's how you get to some of these autonomous operations it's not about removing keystrokes and making it like oh i can type in some command faster it's about you know helping you make decisions much faster in this case you know the decision is you know do i need to go and troubleshoot the network or are things working again the way they're supposed to be working so maybe you should go and look at other things that might be causing um a problem on the retail side like that's a the amount of time that that that saves it's not just the the time it takes to execute commands but it's the time it takes for for people to coordinate across functions that's how you drive a lot of value you know not just into the it or the networking team but into the enterprise at large yeah i've likened the transition to autonomous networking a little bit like the transition to autonomous vehicles right nobody except maybe elon musk would jump into a car with no steering wheel no controls and no driver right but we're all willing to drive a car now and actually like features like parallel park assists and lane change alerts and some of those things and so you know baby steps will get us there and one day down the road we'll have fully autonomous operations but i do think there is some trust that needs to be built up here so a last question for you mike if i if i look at juniper's kind of bigger messaging right juniper talks a lot about experience first networking and uh so can you tell us what that is and then how does abstract fit into that yeah i mean the whole idea behind experience first is this it's the notion that the things we optimize for are changing where it used to be all about you know price and performance it was about you know transport moving packets like that's table stakes right when when you can move the packets the question is what do you do from there it's no longer about is your network up right and we user experience is the new uptime it's not it's just not the case that having you know your high availability is enough to declare success um you've got to look and say well you know are things working the way that you want them to work the applications function in the data center adequately to support the business and if that's the case then you have to do more than you know merely provision things so that they so that connectivity is in place you've got to be able to um assure that that things are working continuously the way they're supposed to you've got to make sure that the applications have um you know not just the the connectivity but they've got you know all the things that the drive overall experience um so for instance what does latency look like within the application um you know are you seeing intermittent drops are you seeing um you know congestion that's leading to to changes in application side behavior um if you look northbound into the application space you know are we seeing things that are happening you know at the application side or maybe within some of the the vmware you know suite of tools and is that leading to the decisions that need to be made or information that needs to be sourced from the underlying network this is like it's it's so much broader than just merely connecting things um the whole role of appstra is essentially saying look first let's let's make it as simple as possible to get the network up and running it's table stakes let's do it well but let's do it quickly so you don't have to spend time on that second you know the whole idea of reliability is not that you know hey i've pushed some configuration out and it should be working it's like can i guarantee can i demonstrate that it's working and that extending those workflows beyond merely pushing the configuration but to looking at sort of the assurance and reliability side that's a big push towards um as you start thinking about you know what are the types of things that impact experience we talked about the policy overlap okay so you've you've set your policy does that mean you're good well maybe but if somebody else has set conflicting policy like isn't the outcome that you want that you've got to secure um that you have a secure posture for a particular set of users whether they're applications or tenants or whoever it's more than merely pushing configuration out it's got to be this closed loop validation that's what apps are trying to do and if we can do that in an environment that's multivenger if we can do an environment that's multi-domain if we can do that in an environment that's multi-technology multi-topology like everything becomes multi over time if abstract can be that unified operations platform over the top of all of those multis like that's what we're trying to drive here at juniper all right well thanks for that mike and uh you know after leading you to a multi-everything world i guess a multi-multi-world so uh you know with that we're gonna wrap up i think that was a great overview of your announcements and you know clearly the data center has never been more important right everything is moving to some kind of data center be it you know a centralized one public cloud or in an edge so uh again uh thanks mike luxon from juniper network for joining me i'm zeus caraval from zcast signing off don't forget to click to subscribe and i'll see you next time

This transcript was generated automatically from the video's captions and may contain errors.

Written By
Zeus Kerravala
Zeus Kerravala
Published: Mar 15, 2022
Updated: Nov 19, 2024
1 minute read
eWeek content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More

I recently spoke with Michael Bushong, Vice President of Cloud-Ready Data Center at Juniper Networks about the changing role of the data center and what it means to the cloud.

Zeus Kerravala

Zeus Kerravala is an eWEEK regular contributor and the founder and principal analyst with ZK Research. He spent 10 years at Yankee Group and prior to that held a number of corporate IT positions. Kerravala is considered one of the top 10 IT analysts in the world by Apollo Research, which evaluated 3,960 technology analysts and their individual press coverage metrics.

eWeek Logo

eWeek has the latest technology news and analysis, buying guides, and product reviews for IT professionals and technology buyers. The site's focus is on innovative solutions and covering in-depth technical content. eWeek stays on the cutting edge of technology news and IT trends through interviews and expert analysis. Gain insight from top innovators and thought leaders in the fields of IT, business, enterprise software, startups, and more.

Property of TechnologyAdvice. © 2026 TechnologyAdvice. All Rights Reserved

Advertiser Disclosure: Some of the products that appear on this site are from companies from which TechnologyAdvice receives compensation. This compensation may impact how and where products appear on this site including, for example, the order in which they appear. TechnologyAdvice does not include all companies or all types of products available in the marketplace.